<JulesIvanicGitte>
> Could you open an issue? Does sound like a bug, but shouldn't be hard to fix
<JulesIvanicGitte>
Will be done ;)
<JulesIvanicGitte>
thanks :)
<headius[m]>
@ju
<headius[m]>
ug
<headius[m]>
Jules Ivanic (Gitter): I'm still not used to Riot
<JulesIvanicGitte>
???
<JulesIvanicGitte>
(is it a joke ? I’m not a native english speaker. I can miss some subtilities)
<headius[m]>
Jules Ivanic (Gitter): Thank you!
<headius[m]>
No sorry, I tried to use @ju to bring up the user completion dialog, but you have to also select a user from that dialog before hitting enter 😆
<headius[m]>
If you can come up with a small reproduction this can be fixed for the 9.2.7 release next week
<headius[m]>
enebo:
<JulesIvanicGitte>
ok
<JulesIvanicGitte>
I’ll try to give you a small open source project reproducing the error tomorrow morning ;)
<headius[m]>
I mean, we shouldn't be making objects with 100 fields no matter what, but it may be rare
<headius[m]>
ah yes, that too
<headius[m]>
After we get past Ruby Kaigi and RailsConf I want to get back to these optimizations and start to tie it all together
<JulesIvanicGitte>
Let me know if you need me to test something
<headius[m]>
Jules Ivanic (Gitter): I'd love to see comparison heap dumps with -Xreify.variables=false and =true (default) if you get a chance
<headius[m]>
you can do that on any version
<JulesIvanicGitte>
I will not a Rubyist for still a long time. I quit my current job at the end of April and then I’ll leave the Ruby world to concentrate on the Scala one
<JulesIvanicGitte>
> I'd love to see comparison heap dumps with -Xreify.variables=false and =true (default) if you get a chance
<JulesIvanicGitte>
Ok. Maybe I can do that tomorrow too
<headius[m]>
Sure, it would be good information for us to see a large app with and without
<headius[m]>
I still suspect turning it off will use more memory, but we shall see
<JulesIvanicGitte>
Plus, now I’m able to produce some “real” traffic on my app while before I was just using 1 API endpoint
<JulesIvanicGitte>
(edited) Plus, now I’m able to produce some “real” traffic on my app while before I was just using 1 API endpoint => I will not a Rubyist for still a long time. I quit my current job at the end of April and then I’ll leave the Ruby world to concentrate myself on the Scala one
<lopex>
headius[m]: if you order a company that comes to visit me, puts me in a box and then unbox me at the place you want, then I might reconsider my travel attitude
<lopex>
I hope it sounds amusing a bit
<lopex>
and, I'll pay for that
<lopex>
or, I guess wont
<headius[m]>
Hah sure...you can rent a cabin on a container ship for a few months
<lopex>
headius[m]: hey, just having my own first jruby production deployment
<lopex>
using my own framework :P
<lopex>
headius[m]: squished docker jruby alpine image with an app to 250MB
<headius[m]>
That's pretty good
<headius[m]>
How much RAM usage?
<lopex>
headius[m]: and all using as/400 env
<lopex>
let me see
<headius[m]>
I want to try to AOT compile JRuby also, probably trying 1.7 first since it's simpler but ideally we'd get to a point where we could AOT both JRuby and the Ruby code together
<lopex>
587M res
<headius[m]>
Not too bad
<lopex>
but JAVA_OPTS=-Xmx2048m -XX:NewRatio=1
<lopex>
and it;s sinatra, not rails based
<lopex>
small one
<headius[m]>
Best kind of Ruby app 😀
<headius[m]>
We were running some Sinatra benchmarks across Ruby impls last week and JRuby was clearly the fastest...maybe you will find some opportunities though.
<lopex>
headius[m]: with the parts I've shown you on gists
<lopex>
headius[m]: the ones that show there's no usable as/400 admin app
<headius[m]>
Haha ok
<lopex>
but seriously
<lopex>
if a job ends then you have to search for the output on spoolfiles
<lopex>
and its not an easy task if you have telnet terminal